Positioning
Yes — but not for the reason usually given. The scarce part of calculus was never the answer; it was the setting up, and that is still the part a tool cannot do for a student. Here is the strongest case on each side, the evidence behind both, and what would change our mind.
“Worth it” needs a subject. Worth it to a seventeen-year-old choosing next year’s courses? To a department deciding what to require? To a teacher deciding what to keep when there is already too much to cover? This page takes the student’s question seriously and refuses the two easy answers: that calculus is timeless and therefore above the argument, and that a machine which computes an integral has therefore made the course pointless. Both are slogans, and a teacher can spot a slogan from the back of the room.
Start with what the tool actually takes away, stated at full strength. If the point of a calculus course is to produce correct derivatives and integrals, that job is largely gone. A symbolic engine will differentiate and integrate faster and more reliably than any student, and a general-purpose assistant will now narrate the steps in plain English. An answer is no longer evidence of anything. When the product is cheap, the course that sold only the product has nothing left to sell.
The exam already treats machine computation as normal, and it did so before the current wave of tools. On the AP Calculus AB exam, a graphing calculator is required on 13 of the 42 multiple-choice questions and on 2 of the 6 free-response questions — 15 of the exam’s 48 questions, where the expected path runs through a machine (checked 19 October 2026; see the sources at the end).
There is an opportunity-cost argument too, and it is the one departments actually make. Hours are finite. Every week spent on techniques of integration is a week not spent on statistics, on data, on modelling, or on the programming that many quantitative jobs now assume. That trade is real, and pretending otherwise weakens the case for calculus rather than strengthening it.
The serious version of the objection is not “nobody needs calculus.” It is that a course which teaches only procedure has lost its reason to exist, and any course that cannot say what it is for will eventually be cut. That is a fair criticism. It is also a criticism of a badly designed course, not of the subject.
Calculus is the language in which rates and accumulations are written down. That sounds like a slogan until you look at what institutions require. MIT’s admissions guidance says it plainly: every other quantitative field at MIT, “from our universally required physics and chemistry courses, to even introductory courses in economics and management science”, depends “on a mastery of the basic principles of calculus as a tool of problem solving” — which is why all of its students, regardless of major, must take or receive credit for both single-variable calculus (18.01) and multivariable calculus (18.02) (checked 19 October 2026).
It is a gateway, and the gateway is real. MIT’s own catalogue describes 18.01A as a review of one-variable calculus “emphasizing material not on the high-school AB syllabus”, and lists its prerequisite as one year of high-school calculus “with a score of 5 on the AB Calculus test”. The AP course is treated as preparation for university calculus, not as a terminal qualification — and the university course it prepares students for is required of everybody.
Then there is the argument that matters most in practice. A wrong answer from a tool is fluent, confident, and formatted exactly like a right one. The person who catches it is the one who can ask whether the sign is possible, whether the size is plausible, and whether the units come out right — and those are calculus habits, not tool habits. That claim is our opinion, but it is the opinion of everyone who has ever marked a plausible-looking wrong answer.
The subject is also being maintained rather than wound down. College Board’s 2026–27 changes to the AP Calculus course description are clarifications, and it states that “course content has not changed”; the number of multiple-choice questions and the timing change from the May 2027 exams. An exam board does not re-time an exam it expects to disappear. (That last sentence is our reading, not College Board’s.)
And access is no longer the argument it once was. OpenStax publishes a complete calculus textbook free in web view and PDF, openly licensed for classroom use. A student who wants to learn the subject no longer needs to buy anything to do it.
Not the subject. The division of labour between the person and the instrument. The tool computes; the person sets the problem up, chooses the model, decides what the answer means, and decides whether to believe it. Those four jobs are what a calculus course is actually for, and they are the four that a multiple-choice answer sheet cannot see.
The assessment reflects that, even now. Of the AP Calculus AB exam’s 190 minutes, 122 — 64% — are sat without a calculator at all: 62 minutes of multiple choice and 60 minutes of free response (checked 19 October 2026). Whatever a classroom allows, the exam still pays for unaided reasoning, and the course overview says what kind: students use “definitions and theorems to build arguments and justify conclusions”.
So the practical question for a teacher is not “should we allow the tool” but “which parts of this unit are the reasoning, and are we grading those?” Three questions do most of the work, and they are the same three whether the answer came from a student, a calculator, or an assistant:
None of that requires new software, a new unit, or a policy statement. It requires asking for the setup and the justification, not just the answer, and being explicit that this is what is being graded.
A good argument should be falsifiable, so here is what would move us. If primary evidence showed that requiring calculus did not predict success in the courses that depend on it, the gateway argument would weaken considerably. If an exam board removed the unaided sections entirely, the assessment argument would go with them. And if the tools became reliably able to judge whether their own output fits the situation — sign, size, units, assumptions — then the verification argument would shrink to almost nothing. None of those is true today, and we would rather say what would change our mind than pretend nothing could.
We use no salary figures and no employment projections, deliberately. Those numbers age badly, they are rarely about calculus specifically, and the argument does not need them. We make no prediction about what AI will be able to do next: statements here about tools are about what we could check today. We name no vendor, product or price, and make no comparison between them. We are not claiming calculus is the right choice for every student — for many, a statistics course is the better use of the year, and saying so costs calculus nothing. And the judgement in the title is editorial opinion; the institutional quotations and the exam facts are sourced and dated, and you can check every one of them in an afternoon.
Keep reading
Pedagogy · 9 min read
If the argument above holds, the practical question is what to allow on Monday. What the public rules actually say, five uses that keep the reasoning with the student, and three traps that quietly take it away.
Free, no spam
Roughly twice a month. Unsubscribe any time.
One email address is all we ask for. New articles, roughly twice a month, and nothing else.